Quote from: deadDMwalking;1087731Now there are metrics like 'take 10' which would be the smart decision if you know the TN, but if you suspect that a 19 isn't likely to be sufficient, you probably wouldn't do it.

Edit -
).

Players should be informed of the target numbers, - unless possibly they're making an opposed check.  They should also know in advance what the consequences of failure are likely to be (this pretty much negates the need for 'fail forward')

Because if there's one thing that ought to characterise being skilled, it's knowing whether that cliff-face actually lies within your capability or not - and also between backing down because it's too risky or continuing with the knowledge that a single mistake might result in a plummet to your death.

Quote from: deadDMwalking;1087731Now there are metrics like 'take 10' which would be the smart decision if you know the TN, but if you suspect that a 19 isn't likely to be sufficient, you probably wouldn't do it.

You can also take 20, if you're not pressed for time / in the middle of combat / otherwise distracted.

Quote from: TJS;1087767Players should be informed of the target numbers, - unless possibly they're making an opposed check.  They should also know in advance what the consequences of failure are likely to be (this pretty much negates the need for 'fail forward')

Because if there's one thing that ought to characterise being skilled, it's knowing whether that cliff-face actually lies within your capability or not - and also between backing down because it's too risky or continuing with the knowledge that a single mistake might result in a plummet to your death.

Why should players know that as opposed to having a really strong hunch? (That is, something short of knowing the underlying numbers with absolute certainty, but having a reasonable guess of, "this is probably beyond a normal human and I could die if I attempt it... but I'm not a normal human.")

EDIT: And that's not saying you're wrong. I'm just curious why players should know these things. (Some systems make that an assumption, others assume just the opposite.)

The main argument I see for always giving specific numbers is, as stated, people who are trained in something have a good idea of what they can or can't do. If you're good enough to have a modifier in climbing, you've climbed enough to judge what is and is not pushing the limits of your ability.

As a two alternate compromise positions, let me suggest;

1) A PC should know when they'd be able to succeed if they took 10 on the check. If they can take 10 its something they're skilled enough at doing that its routine for them. Routine and therefore easy to recognize (the GM could even use the phrase "You've done things like this all the time.").

2) Only give specific DCs to people who are trained in the skill. A talented amateur might be good at climbing up a rocky cliff because of their natural strength, but someone with training can accurately judge just how difficult the climb will be and pick out problems that are beyond their level of skill.

The former is good for a speedy heroic game because it skips the dice rolling when it doesn't matter (routine checks when not under pressure), but keeps the dice rolling for when it matters (pressure situations and things normally beyond your ability) while leaving the more difficult DCs a little more open as to precisely how difficult they are. Its basically just extending the "Passive Perception" rules to general tasks... you notice or successfully perform anything you'd routinely notice or perform and move the game right along.

The latter is good for a campaign where you really want to emphasize the difference between trained and untrained skill use. The untrained skill user has to guess about whether they can take 10 or not (and so might roll and fail when they didn't need to) while the trained skill user just knows if they have to roll for the check or not (effectively chopping the bottom half off the probability curve and replacing it with a 100% line until it can't take 10 on something, then dropping off from there).

I'm cool with only knowing the DC being something the trained PC can do.

in general I don't see a good reason for NOT telling the PCs the DCs.  What's the benefits of hiding it?  Earlier in this thread someone was making the argument that the benefits of a flat distribution is knowing the odds of success.  You can't do that if you don't have the information you need.

More to the point, telling the DC in advance means transparency. It's like making rolls on the table.  If you know the DC in advance everyone knows the GM can't fudge afterwards to get the result they want.  Which is why I said you should also know the result of failure.

If there's a 25% risk of the player falling off a cliff-face to their doom then let's have that out in the open for everyone to be committed to before the Dice are rolled.  That way the player is informed and the GM isn't sitting there scratching there head going "ummm".  And there's no need to 'fail forward'.

I can imagine a few scenarios where there are benefits to hiding the target. Factors that the characters do not know about, for one. (Ongoing magic that has nothing to tip the player's off; however, before throwing out, "shenanigans!" assume the players knew they were going into an area where that was possible, so there's no accusation of pulling something out of the GM's ass just to make life worse for a player.)

Most other scenarios would revolve around keeping tension up instead of devolving the game into a pure math exercise (which is not to say math isn't fun, just that once it's a pure math exercise then it's really about beating the system and not playing the game... unless the game is about beating the system, in which case, awesome! Goal met).

QuoteMore to the point, telling the DC in advance means transparency. It's like making rolls on the table. If you know the DC in advance everyone knows the GM can't fudge afterwards to get the result they want. Which is why I said you should also know the result of failure.

I get that players owe each other and the GM transparency, as they are bound wholly by the rules of the game, but in most games, the GM owes the players no such complete transparency. Some directly state that, others do so indirectly, and only a minority that I can think of affirmatively state that the GM owes transparency back to the players (and most are either what this board would consider storygames OR they are really a board game that has some RPG elements, like Shadows over Hammerhal).

QuoteIf there's a 25% risk of the player falling off a cliff-face to their doom then let's have that out in the open for everyone to be committed to before the Dice are rolled. That way the player is informed and the GM isn't sitting there scratching there head going "ummm". And there's no need to 'fail forward'.

This doesn't solve the GM going, "ummm." It just means that the "ummm" may mean, "well, I didn't expect you to make that kind of a choice, so we're either calling this campaign because you died or I have to go back to the drawing board for a bit," or any number of other answers that are still in "ummm." One other answer could be, "You didn't know there was a friendly golden eagle nearby who saved you from death, but he can't get you up the cliff because you're too heavy!" that makes sense within setting but the GM just pulled out of his ass from an "ummm." Which would also still be a failing forward type mechanic.

Fail forward really isn't connected to this issue. It can happen with or without transparency, and the reasons given to justify fail forward still exist in this model. (Whether or not one feels those issues are valid is a different topic, and approaching saying that people who adjust their dice rolls behind the screen from time-to-time are not actually playing D&D because there's only one true way to do so, which I do not believe you're saying so please do not read that wrong.)

I'm of the camp that the GM should aspire to give consistent descriptions of things in a way that helps the players gauge the difficulty, should be on the watch for confusion on the players' part, and then the players should ask for clarification when they are unsure.  If that's done reasonably well, then the players will have a fair idea of target numbers without knowing exactly *.  On those rare occasions when all that fails, then either something serious happened as a consequence or it didn't.  If it didn't, laugh off the adverse results of the misunderstanding.  (They usually are quite hilarious.)  If it was serious, retcon or otherwise mitigate the consequences as needed.

* This for me is the optimum state of the game, for a variety of reasons.  I'm willing to do the occasional OOC retcon or other fix in order to get it 98%+ of the time with minimal intrusion of the math.  Not least of my reasons is that players good at math will take solid information and turn it into the math in their heads, as part of their calculations of risk.  The players not so good at math will not be as confused, because they'll focus on the problem as it exists from their characters' perspective.

When it comes to "fail forward" or, more accurately, roadblock situations, we basically break it down into life threatening (ex. failed Climb checks) and non-life threatening (ex. failing to notice a secret door).

For non-life threatening roadblocks failure is failure. It's a sandbox world so failing to find the Necromancer's secret lair isn't the end of the campaign; it just means he'll survive and continue to be a threat.

For life threatening ones though we remember that Hit Points are more than meat, but include stamina, morale and luck. So instead of falling to your death on a failed climb check you lose hit points as you become fatigued, lose confidence and push your luck... then try again. You still might fall to your death if you fail enough checks on the climb, but it's not a binary pass/fail anymore so if the players have any common sense they'll figure out the obstacle is beyond their skill level and turn back (I don't require checks to back down the way you came) or, worst case, stay where they are while the other PCs figure out a rescue.

Removing the binary outcome of each check, also means things like climbing or swimming across obstacles where multiple checks are required to cover the distance have more bell curve-like results.

The same applies to our d20-based reaction rolls for social encounters. Each successful check only moves the attitude one step and each additional check uses the original attitude of the subject. Going from wary to helpful takes three checks and failures either increase the DC of subsequent checks (a simple failure... maybe time to quit while you're ahead or at least haven't made it worse) or drop their attitude a stage and increase the DC of further efforts (a failure by 5 or more).

In fact, I actually can't think of many situations in my non-combat mechanics where single d20s are used to completely resolve something of any importance.

Maybe on-the-spot knowledge checks? But even there you automatically know anything a "Take 10" result would give you so any check is just to see if you ever picked up a specific piece of esoteric knowledge beyond the typical and each character in the group can make that check so it's a bell curve from the perspective of what the group knows. Also, failure rarely means anything more than the PCs having to go off what they're presently observing (so you see a big flaming brute, but you don't know what it's called or whether it's vulnerable to cold because it needs heat to survive or is resistant to cold because it's so hot that any cold attack is going to be like an ice cube tossed into a blast furnace).

You can't discount the effects of the PCs generally being in a group too. Sure, a search check to look for clues is a single check (if your passive Perception didn't spot it already), but a party of four is making four checks (if they're bothering to help) to determine the outcome.

The point is... even though the core mechanic is a 1d20, a single d20 roll rarely determines the absolute outcome of situation... which turns the outcome into something closer to what you're looking for with a 2d10 check already.

Quote from: Steven Mitchell;1087961I'm of the camp that the GM should aspire to give consistent descriptions of things in a way that helps the players gauge the difficulty, should be on the watch for confusion on the players' part, and then the players should ask for clarification when they are unsure.  If that's done reasonably well, then the players will have a fair idea of target numbers without knowing exactly *.

This is usually where I tend to fall, but I vary with what the system tells me should be the standard for that system (when it does). So, pretty much in accord with this.

Quote from: Chris24601;1087966When it comes to "fail forward" or, more accurately, roadblock situations, we basically break it down into life threatening (ex. failed Climb checks) and non-life threatening (ex. failing to notice a secret door).

For non-life threatening roadblocks failure is failure. It's a sandbox world so failing to find the Necromancer's secret lair isn't the end of the campaign; it just means he'll survive and continue to be a threat.

I like that description and those two categories. They seem to encapsulate things pretty well. In a sandbox setting, for non-life threatening situations, yeah... fail forward is usually unnecessary because, "what is forward?" And in a more plot-centric setting, it's generally not a good idea to hang non-life threatening situations on a single roll if they are critical to forwarding the game. Unless, of course, you have backups and alternatives (i.e., failing forward... which can include failing utterly in this current task and the plot taking a turn no one expected because of it).

Either way seems to deal with the 1d20 vs. 2d10. If the 1d20-single-check is that critical to keeping the game moving, there may (repeat: may) be a structural issue with the scenario that should be examined or mitigated by the GM. And I'm not sure the answer is moving to a more bell-like distribution because outside-expectation cases are not simply edge cases.

Quote from: Tanin Wulf;1087972Either way seems to deal with the 1d20 vs. 2d10. If the 1d20-single-check is that critical to keeping the game moving, there may (repeat: may) be a structural issue with the scenario that should be examined or mitigated by the GM. And I'm not sure the answer is moving to a more bell-like distribution because outside-expectation cases are not simply edge cases.

I'm pretty sure its not the answer at all. Just because you can only hit the roadblock if you roll a 2 or less, doesn't make 2d10 a cure-all. It just means that only 1-in-100 will hit the roadblock instead of 1-in-10.

As an example from a badly written Living Arcanis mod, the entire adventure was gated behind a ward that could only be bypassed with a dispel magic. The check wasn't horrible. At the level it was designed for you only needed a 10 or better. But it was contingent upon A) having a spellcaster with at least 3rd level spells in the party, B) that spellcaster actually having dispel magic prepared and C) that the spellcaster wouldn't fail his dispel magic checks.

In the case of my group, I had dispel magic prepared (three times actually), but my dice didn't cooperate and so, after blowing two of the three (and presuming there was actually more to the adventure since it was a relatively small tower you run across on a patrol) we moved on. Nope... inside the tower was the adventure so I wasted an entire Origins session on what ended up being a 20 minute session (10 of which was introductory text) with no encounters at all. No XP, no gp, no items... just wandered into the woods, found a tower and because of two bad rolls, we were done.

A bell curve wouldn't have fixed that module. A ward with a secret password that could be guessed from a take-20 search of the surroundings (or an encounter because the guys inside opened the door to come at kill the intruders and it was left open afterwards) would have fixed the module.

Quote from: Chris24601;1087995A bell curve wouldn't have fixed that module. A ward with a secret password that could be guessed from a take-20 search of the surroundings (or an encounter because the guys inside opened the door to come at kill the intruders and it was left open afterwards) would have fixed the module.

A bad module is a bad module - I get that and I'm not disagreeing.

Assuming you needed a 10 or better and we're not otherwise adjusting probabilities by reducing bonuses or what have you there is a difference with the bell curve.

On a d20, rolling 1-9 happens 45% of the time; rolling it twice in sequence happens 20.25% of the time.  If it was a true '50-50', you'd flip heads twice in a row 25% of the time.  

On 2d10, rolling a 9 or lower happens only 36% of the time; rolling it twice in sequence happens only 13% of the time.  

Happening one in eight times is a significant difference than having it happen one in five times - while there are still some people that would get the sucky version of the abbreviated adventure, it's still fewer people overall.

Quote from: TJS;1087909in general I don't see a good reason for NOT telling the PCs the DCs.  What's the benefits of hiding it?  Earlier in this thread someone was making the argument that the benefits of a flat distribution is knowing the odds of success.  You can't do that if you don't have the information you need.

I was the one (or one of the ones) that named more readily discerning odds of success as a benefit of a "flat distribution." However, I didn't mean that players should be able to more easily determine the odds. I'm more interested in the GM being better able to instantly know the odds for the sake of better adjudicating. A lot of the examples listed in this thread where a problem is presented seems to me be a failure to adjudicate appropriately.

That said, I do have a good reason for not wanting players to know the number they need. I don't want them to know the odds. I don't want them "solving" the game like it's a math problem. I want them to face "uncertainty" not just "risk" per Frank Knight's distinction.

In fact, I do like the take 10 and take 20 rules. And I see it as something like this.

-You roll the dice, and the unknown is how good the die roll will be. That's where things could go wrong. This is mathematical risk.
-You take 10, and the unknown is whether or not that will be enough to succeed. Underestimating the challenge is where things could go wrong. This is one form of uncertainty.
-You take 20, and you "roll" as high as you could. Where things could go wrong here is you might not have as much time as you think. There's a much greater chance of being interrupted. That's another form of uncertainty.

You don't ever get guarantees. You just get to pick your poison.

Quote from: Chris24601;1087966For life threatening ones though we remember that Hit Points are more than meat, but include stamina, morale and luck. So instead of falling to your death on a failed climb check you lose hit points as you become fatigued, lose confidence and push your luck... then try again. You still might fall to your death if you fail enough checks on the climb, but it's not a binary pass/fail anymore so if the players have any common sense they'll figure out the obstacle is beyond their skill level and turn back (I don't require checks to back down the way you came) or, worst case, stay where they are while the other PCs figure out a rescue.

Removing the binary outcome of each check, also means things like climbing or swimming across obstacles where multiple checks are required to cover the distance have more bell curve-like results.

Yes. Nothing's ever totally "binary." Until you give up or die, you can always try something else. So even "flat" distributions accumulate into some kind of curve. Even if you do die (or give up) on the first roll, that's all part of what makes the Pareto distribution. Even if the stats are such that you necessarily die on the first roll, all that means is you're on an extreme end of the curve.

But at the same time, you can always claim something is binary. If what I'm interested in knowing is "Do you kill the ogre this round?" or "Do you successfully climb this round," that is a binary, either yes or no. And in the context of the situation, that might actually be what's most important. Maybe it's time sensitive and things hinge on you getting it that round. If that's the case, you can play it out ten thousand times or ten million times with whatever convoluted "curve" dicing mechanic you wish. You'll always have x chance that the answer is "yes" and 1-x that the answer is "no", and only those two points side by side will never form a curve.

Quote from: Chris24601;1087995As an example from a badly written Living Arcanis mod, the entire adventure was gated behind a ward that could only be bypassed with a dispel magic. The check wasn't horrible. At the level it was designed for you only needed a 10 or better. But it was contingent upon A) having a spellcaster with at least 3rd level spells in the party, B) that spellcaster actually having dispel magic prepared and C) that the spellcaster wouldn't fail his dispel magic checks.

I agree with the overall point you're making. But I actually really like it when adventures has an up-front gatekeeper like this.

QuoteIn the case of my group, I had dispel magic prepared (three times actually), but my dice didn't cooperate and so, after blowing two of the three (and presuming there was actually more to the adventure since it was a relatively small tower you run across on a patrol) we moved on. Nope... inside the tower was the adventure so I wasted an entire Origins session on what ended up being a 20 minute session (10 of which was introductory text) with no encounters at all. No XP, no gp, no items... just wandered into the woods, found a tower and because of two bad rolls, we were done.

A bell curve wouldn't have fixed that module. A ward with a secret password that could be guessed from a take-20 search of the surroundings (or an encounter because the guys inside opened the door to come at kill the intruders and it was left open afterwards) would have fixed the module.

It seems to me what was needed is some more urgent reason of why you needed to be in this tower. If you really want to get in, you don't give up after just a couple of bad rolls. Especially not when you had a third dispel magic in reserve.

But also one of the things I probably would have done differently (in running or designing the adventure) is waved the check entirely for this. Not for the sake of saving the adventure or being nice or any of that noise. Just because that was sort of like "the key" to the gate. I mean, look. We're okay with sunlight killing a vampire. Or a blessed crossbow bolt killing a rakshasta. And also we're okay with the flip side. A non-magical, non-silver weapon not harming a werewolf. We don't consider this to be cheating or fudging. It's just how things work. It's part of the game. Not all parts of game are supposed to be shoe-horned into a core mechanic.

I do the same thing with searches. If you say "I search" without being specific enough, I rule it's like finding a needle in a haystack, and count that as automatic failure. If you specify your search and its exactly what, how, and where the thing you're looking for is, then I count that as automatic success. If it's somewhere in between the two extremes, where the outcome isn't so obvious to me, that's when I roll dice. It's a last resort, not a first go-to.

Quote from: deadDMwalking;1088032Assuming you needed a 10 or better and we're not otherwise adjusting probabilities by reducing bonuses or what have you there is a difference with the bell curve.

On a d20, rolling 1-9 happens 45% of the time; rolling it twice in sequence happens 20.25% of the time.  If it was a true '50-50', you'd flip heads twice in a row 25% of the time.  

On 2d10, rolling a 9 or lower happens only 36% of the time; rolling it twice in sequence happens only 13% of the time.  

Happening one in eight times is a significant difference than having it happen one in five times - while there are still some people that would get the sucky version of the abbreviated adventure, it's still fewer people overall.

And what if you needed a 15? Or a 20? The 2d10 system would have you less likely to succeed.

It's almost like there's this thought pattern. Higher Success Rate = More Reliable. Because fewer failures. Bell Curve = More Reliable. Because less random. Since more reliable = more reliable, therefore less random = fewer failures. No one will ever come out and say that. In fact, I expect it to be disavowed instantly. Nonetheless, that's what the tone feels like when we get this sentiment of "bell curves make it all better."

I'm still old school. AD&D 1E. Where a first level fighter, who is considered a veteran, has roughly a 30% chance of success at his core competency. If you could make the dice pull towards the middle, the only thing that would become more consistent would be failure. Everything on that side of the curve does the opposite of most of the alleged benefits of the bell curve. It doesn't even live up to cutting Chris's dilemma in half, much less solving anything.

Quote from: Lunamancer;1088094And what if you needed a 15? Or a 20? The 2d10 system would have you less likely to succeed.

Yes.  The point of a two dice is to discourage an extreme result and encourage a median result.  Effectively, this means characters that can succeed with an average result succeed more often; it also consequently means that if you can only succeed with an extreme result it happens less often.  

Quote from: Lunamancer;1088094It's almost like there's this thought pattern. Higher Success Rate = More Reliable. Because fewer failures. Bell Curve = More Reliable. Because less random. Since more reliable = more reliable, therefore less random = fewer failures. No one will ever come out and say that. In fact, I expect it to be disavowed instantly. Nonetheless, that's what the tone feels like when we get this sentiment of "bell curves make it all better."

Let me say it explicitly.  The PCs are expected to succeed generally.  If your players consistently need a 20 to hit, they're going to be missing a lot whether you use a d20 or 2d10.  Generally, game systems tend to make success for equal foes happen on a 10+ (or 11 if they really want to divide the RNG in equal halves).  In 3.x, two unarmed/unarmored humans might have their Dex bonus to AC and attack; if they are identical twins they hit on a 10+.  Assuming that is the case, rolling on 2d10 increases the odds of success from roughly 1 in 2 to 2 in 3.  People hit more often.

Quote from: Lunamancer;1088094I'm still old school. AD&D 1E. Where a first level fighter, who is considered a veteran, has roughly a 30% chance of success at his core competency. If you could make the dice pull towards the middle, the only thing that would become more consistent would be failure. Everything on that side of the curve does the opposite of most of the alleged benefits of the bell curve. It doesn't even live up to cutting Chris's dilemma in half, much less solving anything.

This is true.  If you design every challenge to require a 15+ to succeed (30% on a d20), you'll decrease your chances of success (21% on 2d10).  You're not explicitly stating it, but you are expecting your players to FAIL at their CORE COMPETENCIES more than they succeed (7 out of 10 times, in fact).  

The reason some people like the distribution curve of 2d10 is it TENDS to align more with our real life expectations.  

When most people do things consistently, they tend to consistently get the same result.  They're not equally likely to have their best or worst day - they tend to have their average.  

That may not be your cup of tea, but I think people on this thread generally understand that decreasing the likelihood of extreme results and increasing the likelihood of average results does exactly that - it makes average results more likely.  That's one reason people roll stats on 3d6 and not 1d20.  You end up with a 90% chance of a 7+ but only a 10% chance of a 15+.  Average people cluster around the average.

Quote from: deadDMwalking;1088100The reason some people like the distribution curve of 2d10 is it TENDS to align more with our real life expectations.

I have no doubt people believe that. Whether it's true is another matter.

QuoteWhen most people do things consistently, they tend to consistently get the same result.

People don't roll dice on everything they do. You'd be surprised how consistent results can be when they don't vary.

QuoteThey're not equally likely to have their best or worst day - they tend to have their average.

When and if people are rolling dice, I assume there will be more than one check during the entire day.

QuoteThat may not be your cup of tea,

 I never said anything about it not being my cup of tea. The OP posed a question, I thought pointing out there are weaknesses to using 2d10 and the benefits aren't actually as advertised was worth mentioning.

Quotebut I think people on this thread generally understand that decreasing the likelihood of extreme results and increasing the likelihood of average results does exactly that - it makes average results more likely.

Well, I can't speak to what people do or don't understand. I can only point out there is a fundamental problem here that transcends opinion. That the number generated by the dice is NOT the result on these sorts of checks. The results are pass or fail. And you can't form a curve out of just two points. The average, I suppose, would be somewhere in the middle, but pass or fail doesn't produce a middle, so there is never an average result to be had. If you want to change that, changing the dice won't do it. You'd have to add at least a third result so you have yes, no, maybe. Then we can talk about whether or not making maybes more common is a good idea and whose cup of tea that might be.

QuoteThat's one reason people roll stats on 3d6 and not 1d20.  You end up with a 90% chance of a 7+ but only a 10% chance of a 15+.  Average people cluster around the average.

And this is perfectly appropriate for generating these sorts of statistics. When I roll strength, I get one of 16 possible results, 3-18. You can analyze the frequencies of the various results and you can indeed find a curve if that is the pattern to the distribution. Whether or not that's desirable, that's something that can be discussed. The point is it's possible.

But if instead the game only had two possible strength ratings: Strong and Weak, you don't get a curve when you analyze those frequencies no matter how you make the determination of who is Strong and who is Weak. Two points don't produce a curve. Whether or not a curve or tendency towards average is desirable is not even on the table. The point is it's impossible. It has nothing to do with liking it or not.

That's the key of it. You need to have at least 3 results before you can even talk about bell curves or tendencies towards average results.

And I'm surprised nobody's called me on the standard check actually having four outcomes, pass, fail, crit, and fumble. But of course we already know the extremes of crit and fumble are far more rare than the more moderate pass or fail. There is a low-res bell curve to it. And on a single d20 roll much less!